Lysine-Directed Conjugation of Ethidium Homodimer to B72.3 Antibody: Retention of Immunoreactivity...

Lysine-Directed Conjugation of Ethidium Homodimer toB72.3 Antibody: Retention of Immunoreactivity

but Altered Tumor TargetingRavi S. Harapanhalli, Khalid Z. Matalka, Peter L. Jones, Ashfaq Mahmood,

S. James Adelstein and Amin I. KassisDEPARTMENT OF RADIOLOGY (NUCLEAR MEDICINE), BRIGHAM AND WOMEN’S HOSPITAL AND

HARVARD MEDICAL SCHOOL, BOSTON, MASSACHUSETTS 02115 USA

ABSTRACT. Ethidium homodimer (EHD) was conjugated to B72.3 monoclonal antibody using a methodwhereby 85–90% of the conjugated EHD remains available for DNA intercalation. Antibody wasthiopropionylated by reaction with N-succinimidyl 3-(2-pyridyldithio)propionate and reduction of pyridyldi-thio groups with dithiothreitol. EHD was maleimido-functionalized with succinimidyl-4-(N-maleimidoethyl)-cyclohexane-1-carboxylate and treated with thiopropionylated antibody to obtain a conjugate containing ;3.4EHD per antibody molecule. For biologic studies, 14C-labeled EHD was synthesized by reductive aminationand conjugated as above. In vitro the conjugate maintained chemical integrity and immunoreactivity, while invivo its targeting of LS174T tumors was reduced compared with that of iodinated antibody. A decrease inisoelectric point of the immunoconjugate was also observed. NUCL MED BIOL 25;3:267–278, 1998. © 1998Elsevier Science Inc.

KEY WORDS. B72.3 antibody, Ethidium homodimer, Biodistribution

INTRODUCTION

Immunoconjugates of monoclonal antibodies (MAbs) with drugs(15–18), toxins (1, 6) and radionuclides (8, 14) have gainedconsiderable attention for the treatment of human cancers. Theunderlying concept is the use of MAbs with their remarkablespecificity and selectivity in targeting tumors as carriers of theintended probes to desired sites. The anticipation that such target-ing will enhance tumor cell killing while minimizing damage tonormal tissues has been shown to be true for various antibodyconjugates, including those containing DNA-binding agents such asadriamycin, doxorubicin (17) and vindesine (15, 18).

In addition to the role antibodies play as agents that enhancetargeting, many of them are internalized by tumor cells. Conse-quently, if such antibodies are conjugated to toxic molecules towhich the plasma membrane is impermeable, they can be used tobring into the cell agents that are otherwise excluded. Such anapproach would enable the assessment of the therapeutic potentialof such molecules.

To this end, we have conjugated to the B72.3 MAb the DNAintercalator ethidium homodimer (EHD), a molecule that cannotpermeate mammalian cell membranes. Although Wagner et al. (20)developed a method for chemically conjugating EHD through itscarbohydrate moiety to proteins such as transferrin, their resultsshowed that the conjugate-bound EHD did not intercalate withDNA, presumably owing to some inadvertent chemical change. Inthe conjugation approach presented here, the protein-bound EHDmolecules retain their ability to intercalate with DNA. However,our data reveal that, although the immunoreactivity and theDNA-binding ability of the B72.3(EHD)3 immunoconjugate were

preserved, its ability to target tumors declined substantially and thatthis appears to have been caused by an inadvertent change in itsisoelectric point (pI).

MATERIALS AND METHODSGeneral

Ethidium homodimer (EHD-I), N-succinimidyl 3-(2-pyridyldithio)-propionate (SPDP) and N-succinimidyl 4-(N-maleimidoethyl)cy-clohexane-1-carboxylate (SMCC) were obtained from MolecularProbes (Eugene, OR). All other chemicals were purchased fromAldrich Chemical Company (Milwaukee, WI). [14C]Formaldehydeat a specific activity of 54 mCi/mmol was obtained from NENResearch Products, DuPont Company (Boston, MA). High perfor-mance liquid chromatographic (HPLC) analyses were carried outon a Waters 510 HPLC system (Milford, MA) equipped with atuneable UV detector. Autogamma 500 (Packard Instruments,Downers Grove, IL) and LS 8000 (Beckman Instruments, Irvine,CA) instruments were used for counting all radioactive samples.Thin-layer chromatographic (TLC) analyses of EHD and its deriv-atives were carried out on silica plates (mobile phase, n-butanol:trifluoroacetic acid:water:acetonitrile:ethyl acetate, 8:0.5:1:16:20).

Production, Purification and Characterization of B72.3Antibody

CULTURING OF B72.3 HYBRIDOMA CELLS. B72.3 hybridoma cells(#F-8950) from the American Type Culture Collection (ATCC,Rockville, MD) were grown in RPMI-1640 medium supplementedper 500 mL with 10% fetal bovine serum (heat-inactivated), 5 mLof 200 mM L-glutamine and 5 mL of 5000 mg/mL penicillin–streptomycin. The hybridomas were recultured twice a week, andthe medium was freshly made every 15 days as L-glutamine degradeswith a 15-day half-life.

Address correspondence to: Amin I. Kassis, Ph.D., 220 Longwood Avenue,Goldenson Building, B242, Boston, Massachusetts 02115; e-mailakassis@hms.harvard.edu

Accepted 15 June 1997.

Nuclear Medicine & Biology, Vol. 25, pp. 267–278, 1998 ISSN 0969-8051/98/$19.00 1 0.00Copyright © 1998 Elsevier Science Inc. PII S0969-8051(97)00203-5

ASCITES PRODUCTION IN MICE. Pristane-primed BALB/c malemice (18–20 g) were obtained from Charles River Laboratories(Wilmington, MA). The hybridomas were injected intraperitone-ally (2 3 106 per mouse in 0.5 mL of phosphate-buffered saline[PBS]). If needed, a second dose of hybridomas was injected 7 dayslater. The ascites was collected on days 18, 22 and 25, and thecombined taps were clarified by centrifugation (10,000 3g) andstored at 220°C. The MAbs were isolated on a protein A/G affinitycolumn (No. 20422, Pierce, Rockford, IL) per the manufacturer’sprocedure. In essence, the fluid was mixed with binding buffer (100mM TRIS, pH ;8) in 1:1 ratio and centrifuged at 10,000 3g toseparate lipoproteins. The supernatant was purified on a proteinA/G column using elution buffer (0.1 M glycine buffer, pH 2–3).The eluate was immediately neutralized, dialyzed against PBS, pH7.4, and concentrated by centrifugal membrane filtration (Centri-con-30, Amicon, Incorporated, Beverly, MA). With this procedure,.95% of the IgG was recovered from the column, and this was then

analyzed by size-exclusion HPLC (Fig. 1) on a BioSep-Sec-S 3000column (7.8 3 600 mm; Phenomenex, Torrence, CA) using 0.1 MPBS, pH 7.2, as an eluant (1 mL/min). The purified IgG MAb wasstored at 220°C in PBS containing 0.1% azide.

RADIOIODINATION OF B72.3. The MAb (200 mg/100 mL PBS) wastransferred to a vial precoated with 10 mg of Iodogen and ;300 mCiof Na125I was added. The closed vial was kept at ambient temper-ature for 10 min with occasional shaking. The reaction was stoppedby transferring the vial contents to a PD-10 Sephadex column(Sigma Chemical Company, St. Louis, MO) which had beenpresaturated with 1% bovine serum albumin (BSA) in PBS. Theeluates (1 mL each) were collected and checked for absorption at280 nm as well as assayed for radioactivity in a dose calibrator. Theprotein eluting in the void volume accounted for 88% of theradioactivity. It was pooled and concentrated on a Centricon-30and analyzed on instant TLC plates (ITLC SG, Gelman Sciences

FIG. 1. BioSep-Sec-S 3000 size-exclusion HPLC profile. (A) Crude B72.3 ascites. Note two major peaks corresponding tomolecules with molecular weight of albumin and IgG (shown by arrow). (B) Purified IgG peak.

268 R. S. Harapanhalli et al.

Incorporated, Ann Arbor, MI) with PBS as the mobile phase. Thepresence of $98% radioactivity at the origin indicated the purity ofthis radiochemical. The specific activity of 2.5 mCi/mg of IgGcorresponded to an 125I:IgG ratio of 1:6.

ASSESSMENT OF IMMUNOREACTIVITY OF 125I-B72.3 BY DIRECT RADIO-

IMMUNOASSAY. Since B72.3 is known to react strongly with bovinesubmaxillary mucin (BSM), the immunoreactivity of the 125I-B72.3was assessed in 96-well microtiter plates (Costar, Cambridge, MA)coated with 1 mg BSM (4). In brief, serial dilutions of 125I-B72.3 (15mg, 7.5 mg, etc.), prepared in 0.1 M PBS containing 1% BSA and0.1% Tween 80, were added to each well and the plates wereincubated at 37°C for 1 h. The plates were then washed five timeswith PBS in a plate washer (model 1250 Immunowash, Bio-RadLaboratories, Incorporated, Hercules, CA), and the radioactivityassociated with each well was counted on a g-counter. The datawere analyzed and are presented as a scatchard plot (Fig. 2).

Synthesis of [N-Methyl-14C]EHD from[14C]Formaldehyde

EHD (1 mg, 1.17 mmol) was dissolved in 200 mL of methanol, and250 mCi of an aqueous solution of [14C]HCHO1 (10 mL, 54mCi/mmol) was added (Fig. 3). After 30 min at room temperature(RT), 200 mL of bicarbonate buffer (0.1 M bicarbonate, 0.5 MNaCl, pH 8.3), 63 mL of aqueous sodium cyanoborohydride (10mg/mL, 3.4 mmol) and 200 mL of methanol were added. The tightlyclosed vial was kept at RT for 48 h and monitored by radio-ITLC inPBS, wherein free formaldehyde and its oxidation products movewith the solvent front and EHD remains at the origin. The product(52% yield) was purified by preparative ITLC in PBS. The stripcontaining EHD was eluted with n-butanol:trifluoroacetic acid:water:methyl cyanide:ethyl acetate (8:1:1:16:20) and evaporated todryness by nitrogen purge. Three coevaporations with dry acetoni-trile ensured the product in pure form, free of any organics andmoisture.

Formation and Characterization of the EHD-B72.3Conjugate

SYNTHESIS.2 A gently stirred solution of 16 mg of B72.3 in 4 mLof PBS was treated with slow addition of 1 mg of SPDP in 150 mLof ethanol at RT for 40 min (Fig. 4); unreacted SPDP was removedby gel filtration on a PD-10 column, and the eluate was concen-trated on a Centricon-30. To determine the number of conjugated2-pyridyl disulfide groups (3), we subjected the immunoconjugate toa buffer exchange with 0.1 M sodium acetate buffer, pH 4.4,containing 0.1 M NaCl using Centricon-30.

To a portion of diluted protein solution at RT, 100 mL of 50 mMdithiothreitol (DTT) was added, and absorption at 343 nm and 280nm was assessed on a UV-VIS spectrophotometer after 30 min.Based on the observed OD343 of the solution and the molarabsorptivity of 2-thiopyridone (2-PT)ε343 5 8080, the concentra-tion of 2-PT released by DTT reduction from the MAb wasobtained. Knowing the ε280 of 2-PT (5100), the OD280 of theprotein 5 total OD280 2 [concentration 2-PT 3 5100]. To themain stock, 10 mg of DTT was added and incubated for 45 min atRT. The thiolated antibody was purified on a PD-10 columnpre-equilibrated with PBS, and the protein eluates were concen-trated (vide infra) and stored under argon.

To 1 mg of EHD (1.17 mmol), a solution of one equivalent (0.143mg, 1.17 mmol) of 4-dimethylaminopyridine (DMAP) in 25 mL ofdry dimethylformamide (DMF) was added. The mixture was vor-texed and SMCC (0.39 mg, 1.17 mmol) in 25 mL of dry DMF wasadded, the vial tightly capped and vortexed again, and the samplewas incubated in the dark at RT overnight. TLC analysis revealeda sharp spot (Rf 5 0.41) above the EHD spot (Rf 5 0.34) thataccounted for more than 85% of the fluorescence absorption. PBS(1 mL) was added, and the solution was extracted with ethyl acetate(3 3 1 mL) to remove unreacted SMCC and DMAP. TheEHD-containing aqueous phase was purged with argon and added tothe thiopropionylated MAb solution in PBS (pH 7.4) under argon.The mixture was gently vortex-mixed and incubated in the dark atRT overnight; 2 mg of N-ethylmaleimide in 50 mL of DMF was thenadded and the mixture was incubated for 1 h. Dialysis against PBS(3 3 0.5 L) resulted in a bright-red clear solution. Based on the ε490

of EHD (8.34) and the observed OD490 for the conjugate, theconcentration of bound EHD was determined. Knowing the ε280 ofEHD (66.94), we calculated its contribution to the absorptivity atOD280. Based on the OD280 for protein alone, the protein concen-tration could also be computed. Routinely three to four EHDmolecules were conjugated per molecule of protein.

ASSESSMENT OF THE INTEGRITY OF B72.3(EHD)3 BY SIZE-EXCLUSION

HPLC. A BioSep-Sec-S 3000 size-exclusion HPLC column (600 37.8 mm) was equilibrated with phosphate buffer (0.1 M, pH 7.2, 1mL/min) and calibrated by injecting protein standards of varyingmolecular weights (Bio-Rad Laboratories, Incorporated): bovinethyroglobulin (MW 670 kDa), bovine g-globulin (MW 158 kDa),chicken ovalbumin (MW 44 kDa), horse myoglobin (MW 17 kDa)and vitamin B12 (MW 1350 Da). A calibration plot of retentiontime versus molecular weight was constructed. The retention timesof B72.3 antibody, B72.3(EHD)3 and 14C-labeled immunoconju-gate were noted, and from the calibration plot the molecularweights of native and modified antibody were calculated. For the

1 The purity of commercially procured [14C]formaldehyde was determined byconverting it to a precipitable dimedon adduct and estimating the radioac-tivity (8). A purity of 89% was found.

2 The presence of any dissolved oxygen is undesirable during all chemicalmanipulations involving a thiolated antibody. Therefore, all buffers weredegassed in vacuo and purged with argon. It was also important to maintainan inert atmosphere during the chemical reactions.

FIG. 2. Scatchard plot of radioimmunoassay of 125I-B72.3tested against BSM. Note high association constant for puri-fied B72.3 antibody. Dotted line is linear fit through data.

Ethidium Homodimer–Antibody Conjugate 269

construction of a radio-HPLC profile, the eluates from 14C-labeledimmunoconjugate were collected every minute and the activitycounted in a scintillation counter using Aquasol (20 mL).

ASSESSMENT OF IMMUNOREACTIVITY OF B72.3(EHD)3 BY INDIRECT

RADIOIMMUNOASSAY. The procedure of Carney et al. (4) developedfor ELISA was adapted with some modifications for radioimmuno-assay (RIA). The BSM-coated plates were prepared as describedabove. Stock solutions of B72.3 and B72.3(EHD)3 were prepared at10 mg/mL and were serially diluted down to 0.1 mg/mL in PBScontaining 1% BSA and 0.1% Tween 80. A 100-mL aliquot of theantibody solution was added, and the plates were incubated at 37°Cfor 1 h and then washed five times with PBS in a microplate washer.125I-rabbit antimouse IgG (NEN Research Products, ;150,000cpm/100 mL/well) prepared in the same dilution buffer was added,the plates were incubated for 1 h at 37°C and washed five timeswith PBS in the plate washer before counting.

Determination of pI of MAb by Isoelectric Focusing

The electrophoretic mobilities of proteins were determined in anisoelectric focusing chamber (EC-Apparatus Corporation, St. Pe-tersburg, FL) equipped with a thermal control unit. The sampleswere focused on a 1% agarose focusing gel (pI 3–10; Iso-Gel, FMCBioproducts, Rockland, ME) that was prefocused for 10 min at 15°C

at a constant voltage (500 V) with the current and power set at 20mA and 10 W, respectively. The samples (15 mg/10 mL) wereloaded at the cathode pole (1 M NaOH) along with pI markers(FMC Bioproducts, Rockland, ME). After focusing at 15°C for 90min, the gels were fixed for 20 min in a solution containing 18 g ofsulfosalicylic acid, 30 g of trichloroacetic acid and 180 mL ofmethanol diluted to 500 mL with water and then dried, washed andstained with Coomassie brilliant blue R-250.

Preparation of LS174T Tumor-Bearing Mice

All experiments were carried out in compliance with the NationalInstitutes of Health standards and the Public Health Service policyon the use of laboratory animals. Human colonic adenocarcinomaLS174T cells (#F-11130, ATCC) were maintained as monolayers inDulbecco’s modified Eagle’s medium (GIBCO BRL, Life Technol-ogies, Grand Island, NY) containing 10% fetal bovine serum (50mL total) and enriched with 100 mM minimum nonessential aminoacids, 2 mM L-glutamine and 50,000 U/L of penicillin–streptomy-cin. The cell cultures were passaged twice a week. To providesingle-cell suspensions for injection, the cells were gently pipettedand repeatedly passed through a syringe with a 22-gauge needle.Athymic nu/nu female mice 4–5 weeks old (Harlan SpragueDawley, Indianapolis, IN) were injected subcutaneously in the flank

FIG. 3. Schematic for radiolabeling of ethidium homodimer with 14C. Reaction I demonstrates effect of pH on undesirablereduction of cyclic Schiff-base structure. In reaction II, note that at pH 8.8 internal Schiff-base structure tautomerizes topseudobase form and, following labeling reaction, it reverts back to its original form upon acidification.

270 R. S. Harapanhalli et al.

with 1 3 106 cells/0.1 mL PBS/mouse. With a doubling time ofapproximately 2 days, tumors were palpable within 4 days, and byday 8 they had reached an average weight of ;0.3 g (;7.5-mmdiameter).

Biodistribution of 125I-B72.3, B72.3[14C-EHD]3, and125I-B72.3(EHD)3

PHARMACOKINETICS. For pharmacokinetic experiments mice re-ceived an intravenous injection (lateral tail vein) of 10–20 mg ofthe labeled antibody and were sacrificed between 1 min and 5 dayslater. The tissue activity was counted in either a g-counter or aliquid scintillation counter. Based on the fractional tissue activity,the percent activity per organ and the tumor-to-nontumor (T/NT)ratios were determined.

SAMPLE PREPARATION FOR LIQUID SCINTILLATION COUNTING. Areliable and reproducible method of sample preparation and quenchcorrection for the loss in counting efficiency was needed. Scintil-lation vials with plastic cap-liners were used. The tissue sampleswere rinsed with water and blotted, and each was placed in a vial(20-mL capacity) and weighed. Following the addition of NCS-II

(Amersham Corporation, Arlington Height, IL; 1 mL/0.1 g tissue),the vials were incubated overnight on an orbital shaker at 37°Cuntil the solution cleared. A freshly prepared solution of benzoylperoxide in toluene (1 g/5 mL; 0.3 mL/mL of NCS-II used) wasadded and the samples vortex-mixed and kept at 45°C for 1 h untildiscoloration was completed. To maximize counting efficiency andminimize background, the samples were neutralized with glacialacetic acid (30 mL/mL of NCS-II used), BCS-NA (Amersham, 15mL) was added and, to prevent any phase separation, Tween 80(0.1–1 mL/vial) was added prior to determining radioactive con-tents in a liquid scintillation counter. Quench correction wascarried out by constructing a calibration curve (counting efficiencyversus added counts per minute) after spiking each sample with aknown amount of 3H.

RESULTS

In a typical HPLC profile (UV280) of ascites (Fig. 1A), two majorpeaks with molecular weights corresponding to IgG (150 kDa, 19.2min, pointed by an arrow) and albumin (60 kDa, 21.3 min, majorpeak) can be seen on the chromatogram. After purification of IgGfrom the ascites on a protein A/G column and dialysis, the HPLC

FIG. 4. Schematic for conjugation of EHD to B72.3. Presence of intrinsic disulfide links in protein and their stability towardchemical reagents are shown.

Ethidium Homodimer–Antibody Conjugate 271

FIG. 5. Size-exclusion HPLC profile of B72.3[14C-EHD]3 conjugate. (A) Monitored by UV at 280 nm. (B) Monitored by liquidscintillation counting. Note chemical and radiochemical homogeneity of immunoconjugate. Also note that there was 1-mindelay in measurement of radioactivity from eluates.

272 R. S. Harapanhalli et al.

profile (Fig. 1B) demonstrates the presence of a single 150 kDamolecular weight species (19.2 min). Its immunoaffinity is evidentfrom the Scatchard analysis (Fig. 2; Ka 5 1.4 3 1010 M21) of itsbinding to antigen BSM and is somewhat higher than the valuereported using LS174T antigen TAG-72 (10, 11).

In the synthesis of [N-methyl-14C]EHD from [14C]formaldehydeby reductive amination (Fig. 3), if the Schiff-base form of EHD isnot protected by conversion to its pseudobase, a tetrahydro-N-methyl EHD is formed as evidenced by the change in fluorescencefrom red to blue (reaction I). However, reacting the pseudobase ofEHD with formaldehyde prior to its back conversion to theSchiff-base form (reaction II) bypasses this problem. The labelingresulted in a specific activity of 13.5 mCi/mmol from [14C]HCHO(54 mCi/mmol), indicating that, on average, approximately twomolecules of formaldehyde had reacted per molecule of EHD.Radio-TLC showed a single highly fluorescent spot (Rf 5 0.34) thataccounted for .98% of the 14C activity.

In the procedure for conjugation of EHD to B72.3 (Fig. 4), thethiol groups were introduced by the method of Carlsson et al. (3)using SPDP, a heterobifunctional cross-linking agent that reactswith ε-amino groups of lysine residues in a nonsite-specific manner.Pyridyldithio groups thus introduced (6–8/IgG) were reduced withDTT at pH 4.4 (acetate–EDTA) wherein only the extrinsicdisulfides were prone to reduction. EHD was made thiol-reactive bythe introduction of maleimide groups upon reaction with theheterobifunctional agent SMCC. The thiolated antibody reactedwith this chemical in oxygen-free PBS (RT, overnight, argon)yielding a conjugate containing three to four EHD molecules. In thesame manner [N-methyl-14C]EHD was conjugated to B72.3. In thesize-exclusion HPLC (UV/radio) of this radioimmunoconjugate(Fig. 5), a small shoulder at the base prior to the IgG peak may beseen because of the dimeric aggregate (;10–15%) formed duringconjugation. Chemical and radiochemical homogeneity of thepreparation are confirmed by the observation that all injectedactivity was accounted for in the IgG peak. A plot of calibration ofthe HPLC column with various protein standards revealed a goodcorrelation between protein molecular weights and retention times(data not shown). The specific activity of the immunoconjugate was

FIG. 6. Indirect RIA of B72.3 andB72.3(EHD)3 against BSM using125I-rabbit antimouse IgG. Anti-body concentration at half-maxi-mum for both proteins is around 1mg/mL.

FIG. 7. Isoelectric focusing gel profile of native B72.3,B72.3(EHD)3 and pI markers. Note presence of three iso-forms of native B72.3 (entries 1 and 3) and of several isoformsof B72.3(EHD)3 (entry 4).

Ethidium Homodimer–Antibody Conjugate 273

353.8 mCi/mmol; a 3-fold increase from 113.5 mCi/mmol for[N-methyl-14C]EHD is consistent with the conjugation ratio of 3:1determined by spectrophotometry. Based on fluorescence titrationstudies, the ability of protein-bound EHD to intercalate DNA wasfound to be nearly 90% (data not shown). RIA of B72.3(EHD)3

(Fig. 6) indicates that conjugation of three to four EHD per B72.3molecule by the SPDP–SMCC method had no effect on itsimmunoreactivity. The isoelectric focusing data (Fig. 7) show thatthe native antibody consists of three close moving isoforms (pIs of7.0, 7.1 and 7.2) with a midpoint pI of 7.1 (lanes 1 and 3). Theconjugate (lane 4), on the other hand, has a profile that containsseveral isoforms of decreased pI from 5.9 to 7.0 (midpoint pI 6.4).The pI of these antibody preparations was deduced from thecalibration plot of the pI of the standard markers versus the distancetraveled from the cathode (Fig. 8). The marker proteins resolvedinto distinct bands representing the proteins of varying electro-phoretic mobilities (Fig. 7, lanes 2 and 5; pI range 3.6–10.2), anddiluting the markers in 10 mL of PBS (equaling that for theantibody samples) did not affect their mobilities (lane 5).

The tumor uptake of 125I-B72.3 (Fig. 9) increased with time,reaching 28–30% of injected dose per gram organ (ID/g) by 24 h,and the activity remained close to these levels thereafter. Theradioactivity was not retained in normal tissues and maximumtumor-to-nontumor ratios were seen at 72 h. However, the biodis-tribution profile of B72.3[14C-EHD]3 (Fig. 10A) reveals a tumoruptake of 7–9% by 12 h that remained steady. This is nearly3-to-3.5-fold less than the uptake of unconjugated 125I-B72.3.Concurrently, a high percentage of activity was found in the liver(40–60% ID/g), spleen (10–15% ID/g), and kidneys (5–8% ID/g).The radiolabeled 125I-B72.3(EHD)3 shows a similar profile (Fig.

10B) in that tumor activity gradually rose to 8–9% and remainedunchanged thereafter. However, its localization within the liver by1 h (;20% ID/g) was nearly 2.5-fold less than that of the[14C-EHD]-labeled MAb, and this activity declined thereafter. Aconcurrent rise of activity in the thyroid was also noted with time.The blood pharmacokinetics of all three agents are summarized inTable 1. While the blood clearance of all preparations was biphasic,the clearance of conjugates was nearly twice as rapid as that of thenative antibody.

DISCUSSION

[14C]-formaldehyde (2, 7, 12) was reductively methylated toethidium homodimer (Fig. 3). Initially labeling at pH 5–6 resultedin a change of the native fluorescence from red to blue, and thesilica TLC analysis showed that compared to the Rf of EHD (0.55),the newly formed product had a much higher value of 0.8. Althoughprimary amines are converted into secondary amines during reduc-tive amination, this is not likely to modify the fluorescenceproperties of EHD. The change in color meant the native chro-mophore had been modified. The two ring nitrogens in the form ofcyclic Schiff bases (i.e., C 5 N1R1R2), an essential part of thechromophore, may have been reduced inadvertently during reduc-tive amination with cyanoborohydride, and this may have led to thefluorescence color change. To avoid this, the vulnerable chro-mophore was converted to its pseudobase form under alkalineconditions, reductive amination was carried out, and the originalred fluorescence was regenerated by acidification. A radiochemicalyield of ;52% and purity of $98% were achieved following

FIG. 8. Calibration plot of pI of known proteins (entries 2 and 5, Fig. 7) versus distance (cm) migrated from cathode pole.

274 R. S. Harapanhalli et al.

purification by preparative ITLC. In essence, a facile and mildmethod of labeling EHD with 14C has been developed.

In our early attempts to conjugate EHD to B72.3, the maleimido-functionalization of EHD was carried out in PBS using stoichiomet-ric amounts of SMCC. The isolated yield of the immunoconjugatewas only ;25%, the remaining being aggregated protein. Realizingthat this was due to the lack of solubility of SMCC in aqueousbuffers, we then carried out the reaction of EHD with SMCC inDMF, and any unreacted SMCC was extracted out of the solutionwith ethyl acetate before adding the maleimido-functionalized EHDto the thiopropionylated antibody. This modification improved theyield of immunoconjugate to 60% and emphasizes the need to avoidformation of aggregates associated with the use of SMCC in aqueoussolutions.

The biodistribution results of 125I-B72.3 and B72.3[14C-EHD]3immunoconjugate indicate that while the immunoreactivity of theEHD conjugate was similar to the native iodinated MAb (Fig. 6), itstumor uptake was 3- to 3.5-fold less (7–9% versus 28–30% ID/g).Since the injected activity was recovered in the IgG peak (Fig. 5),this decrease is not due to aggregation of the conjugate as demon-strated by the size-exclusion HPLC analysis of the conjugates.Interestingly, our earlier work (19) had demonstrated that while theradioiodination of the anti-insulinoma MAb A1D2 at iodine:protein ratios of 15:1 and 25:1 minimally affected in vitro immuno-reactivity assessed by RIA, it nevertheless led to an ;3-foldreduction in tumor uptake in vivo; in fact, the higher the iodine:MAb ratio, the lower the tumor uptake. It is clear, therefore, that

such in vitro testing of immunoconjugates does not necessarilypredict their in vivo behavior. Other investigators have also reportedsuch discrepancies between in vitro immunoassay and in vivobiodistribution results. For example, Sakahara et al. (13), assessingthe in vitro affinity of two MAb undergoing identical labelingprocedures and of the same MAb radiolabeled with two differentisotopes, found that the data did not reflect the in vivo behavior ofthese conjugates. Rodwell et al. (9) compared the in vivo behaviorof R9.75 MAb iodinated by either conventional tyrosine-directedelectrophilic labeling or an iodinated tyrosine-containing peptidesite-specifically attached to the oxidized oligosaccharides present onthe Fc region of the antibody. In these studies, the site-specificallyradiolabeled antibody localized in tumors with an 18-fold greaterefficiency than the corresponding conjugate modified nonselec-tively on its tyrosines, despite the fact that in vitro both radio-immunoconjugates had comparable binding properties with theantigen.

To address whether this difference in tumor uptake was aconsequence of the dissociation of the 14C label, biodistributionexperiments were also performed using an 125I-B72.3(EHD)3 con-jugate. A low tumor uptake (8.5% ID/g) corroborated the aboveresults. Interestingly, the liver uptake of 125I-B72.3(EHD)3 was notso high as that seen with [14C-EHD]-labeled immunoconjugate.However, since high radioactivity (125I) was seen in the neckcontents and urine, presumably owing to free iodide, the data seemto indicate that while both immunoconjugates were rapidly takenup and retained by the liver, the radioiodine became dissociated

FIG. 9. Biodistribution profile of 125I-B72.3 in LS174T tumor-bearing athymic nude mice. Note that tumor uptake remainedsteady for up to 72 h after injection and then cleared at very slow rate. Maximum T/NT ratio was seen at 72 h (n 5 5 andstandard deviations between 5–11% of mean). Abbreviations: LV, liver; SP, spleen; K, kidney; SI, small intestine; LI, largeintestine; ST, stomach; LU, lungs; MS, muscle; SKT, skeleton; H, heart; SKN, skin; BA, bladder; NK, neck contents; BL,blood; U, urine; TU, LS174T tumor.

Ethidium Homodimer–Antibody Conjugate 275

FIG. 10. Biodistribution profile of B72.3(EHD)3 immunoconjugate. (A) Labeled with 14C. (B) Labeled with 125I. Low tumoruptake (approximately 7–9%) may be noted for both immunoconjugates (n 5 5 and standard deviations between 6–11% ofmean). Abbreviations: LV, liver; SP, spleen; K, kidney; SI, small intestine; LI, large intestine; ST, stomach; LU, lungs; MS,muscle; SKT, skeleton; H, heart; SKN, skin; BA, bladder; NK, neck contents; BL, blood; U, urine; TU, LS174T tumor.

from its immunoconjugate (probably by the action of various liverdehalogenases) and was then either excreted or taken up by thethyroid. This argument is supported by various studies demonstrat-ing that the tyrosine-directed iodination of MAb usually leads to invivo deiodination and, therefore, a high thyroid activity. Forexample, using 125I-81C6 MAb in tumor (D-54 MG)-xenograftednude mice, Zalutsky et al. (21) reported that the antibody directlyiodinated by Iodogen resulted in high thyroid uptake and increasedurinary excretion of radioiodine as compared to the same antibodylabeled with a small radioiodinated molecule. It is probable thatsuch in vivo deiodination is due to the structural similarity betweenthe iodotyrosines created in conventional protein iodinations andthyroid hormones, compounds known to undergo extensive enzy-matic dehalogenation (5).

Since the biodistribution experiments (Figs. 9 and 10) demon-strate a significant reduction in the tumor-targeting capacity of theB72.3(EHD)3 that was due to neither alteration in the immunore-activity nor molecular weight changes of MAb (Figs. 5 and 6), thepIs of these immunoconjugates were assessed. The midpoint pI ofthe native B72.3 MAb decreased from 7.1 to 6.4 following EHDconjugation (Figs. 7 and 8). The conjugate also showed a series ofclosely related bands, indicating a heterogeneity associated with theformation of several isoforms during such lysine-directed conjuga-tion. Such a decrease in pIs could indicate either the acquisition ofnegative charges by the antibody or the neutralization of its nativepositive charges during the following three phases of conjugation.(1) SPDP–antibody conjugate. Since the SPDP–SMCC-based con-jugation method directs the EHD molecules to the ε-amino groupsof the lysine residues on the antibody, these amino groups will beconverted into amides, and for every SPDP molecule conjugated,one positive charge will be neutralized. In our studies, six to eightSPDP molecules were conjugated on average per MAb moleculeduring its thiopropionylation and, consequently, six to eight lysylresidues were neutralized, an overall loss of six to eight positivecharges. (2) The EHD molecule. Only one of the eight nitrogenatoms in this molecule was used in its conjugation to the antibody.Thus, the MAb can potentially carry seven positive charges perconjugated EHD molecule. However, the pKa of the amino groupsof EHD differ greatly and, as such, all of them will not be positivelycharged at physiologic pH. For example, the pKa of the fouraromatic amino groups is ;4.6, and at pH 7.4 they will lose a protonand hence their positive charge. In addition, the two nitrogenatoms within the phenanthridine rings of EHD in the form of Schiffbases, each carrying a single positive charge, could revert to thepseudobase form and also lose their charge at pH 7.4. Because thetwo amino groups in the aliphatic side chain of EHD are the mostreactive, it is highly likely that either one of them could interactwith the SMCC cross-linking reagent during the conjugationreaction and, therefore, lose its charge when an amide link isformed. Since the pKa of such amino groups is in the range of 9–10,the remaining aliphatic amino group would be expected to retain itspositive charge at physiologic pH of 7.4 and contribute a single

positive charge to the MAb–EHD immunoconjugate. (3) MAb–EHD conjugate. As mentioned above, the conjugation of six to eightSPDP molecules per MAb leads to an overall loss of an equalnumber of positive charges. In addition, a single positive charge iscontributed to each MAb molecule per EHD molecule conjugated.Since three to four EHD molecules were conjugated per MAb in ourstudies, the change in the antibody pI should therefore reflect anoverall loss of three to four positive charges. The observed decreaseof the MAb pI from 7.1 to 6.4 (Figs. 7 and 8) agrees with theseexpectations. Both the differences in the blood pharmacokinetics of125I-B72.3 and 125I-B72.3(EHD)3 (Table 1) and their biodistribu-tion profiles (Figs. 9 and 10) appear, therefore, to be related to thedecrease in pI of the EHD-labeled antibody, although it should bementioned that changes in hydrophobicity due to conjugation ofsmall molecules such as EHD may as well be responsible in part.

As mentioned earlier, tumor localization of the EHD–immuno-conjugate was reduced severalfold compared with 125I-B72.3 MAb.In normal tissues, however, both the uptake and retention differed(Figs. 9 and 10). For example, whereas the localization of theiodinated MAb within the liver was ;20% within the first fewminutes after injection and declined rapidly with time, ;40% of theMAb–EHD conjugate activity was found in the liver by 1 min andthe radioactivity increased to ;60% over the next 3 days. Thealtered pI may have led to the recognition and rapid uptake of theMAb by the reticuloendothelial system.

The results herein have confirmed that an in vitro immunoreac-tivity assay is not an indication of the in vivo behavior of immuno-conjugates. They address the need for determining the extent oftumor targeting of immunoconjugates of drugs and toxins prior totheir use in cancer therapy experiments. Owing to the limitations ofimmunoconjugate prepared by lysine-directed conjugation chemis-try, it would be fortuitous to conjugate EHD to B72.3 MAb in amanner producing an immunoconjugate with unaltered pI thatwould target tumors efficiently (;30% ID/g). We are presentlyexploring site-specific conjugation of EHD to the MAb via thecarbohydrates of the Fc region with the objective of overcoming thedrawbacks in the current procedure and of developing pretargetingapproaches for radioimmunotherapy.

This work was supported by a grant from Boston Life Sciences,Incorporated, Waltham, MA. R. S. H. is recipient of a NationalResearch Science Award.

References1. Blair A. H. and Ghose T. I. (1983) Linkage of cytotoxic agents to

immunoglobulins. J. Immunol. Methods 59, 129–143.2. Borch R. F., Bernstein M. D. and Durst H. D. (1971) The cyanohydri-

doborate anion as a selective reducing agent. J. Am. Chem. Soc. 93,2897–2904.

3. Carlsson J., Drevin H. and Axen R. (1978) Protein thiolation andreversible protein–protein conjugation: N-succinimidyl 3-(2-pyridyldi-thio)propionate, a new heterobifunctional reagent. Biochem. J. 173,723–737.

TABLE 1. Blood Pharmacokinetics of 125I-B72.3, B72.3[14C-EHD]3, and 125I-B72.3(EHD)3

Antibody

t1/2 (h) %ID/g

a b 1 min 1 h 6–9 h 24–25 h 48–50 h 69–77 h125I-B72.3 2.3 57 69.2 ND 29.5 21.5 14.5 9.1B72.3[14C-EHD]3 1 32 23.6 12.9 ND 4.1 2.71 2.4125I-B72.3(EHD)3 1 35 38.0 19.0 8.2 5.6 5.0 3.1

Ethidium Homodimer–Antibody Conjugate 277

4. Carney P. L., Rogers P. E. and Johnson D. K. (1989) Dual isotope studyof iodine-125 and indium-111-labeled antibody in athymic mice.J. Nucl. Med. 30, 374–384.

5. Eary J. F., Krohn K. A., Kishore R. and Nelp W. B. (1989) Radiochem-istry of halogenated antibodies. In: Antibodies in Radiodiagnosis andTherapy (Edited by Zalutsky M. R.), pp. 83–102. CRC Press, Incorpo-rated, Boca Raton, Florida.

6. Ghose T. I., Blair A. H. and Kulkarni P. N. (1983) Preparation ofantibody-linked cytotoxic agents. Methods Enzymol. 93, 280–333.

7. Jentoft N. and Dearborn D. G. (1979) Labeling of proteins by reductivemethylation using sodium cyanoborohydride. J. Biol. Chem. 254,4359–4365.

8. Macklis R. M., Kinsey B. M., Kassis A. I., Ferrara J. L. M., Atcher R. W.,Hines J. J., Coleman C. N., Adelstein S. J. and Burakoff S. J. (1988)Radioimmunotherapy with alpha-particle-emitting immunoconjugates.Science 240, 1024–1026.

9. Rodwell J. D., Alvarez V. L., Lee C., Lopes A. D., Goers J. W. F., KingH. D., Powsner H. J. and McKearn T. J. (1986) Site-specific covalentmodification of monoclonal antibodies: In vitro and in vivo evaluations.Proc. Natl. Acad. Sci. USA 83, 2632–2636.

10. Roselli M., Schlom J., Gansow O. A., Brechbiel M. W., Mirzadeh S.,Pippin C. G., Milenic D. E. and Colcher D. (1991) Comparativebiodistribution studies of DTPA-derivative bifunctional chelates forradiometal labeled monoclonal antibodies. Nucl. Med. Biol. 18, 389–394.

11. Roselli M., Schlom J., Gansow O. A., Raubitschek A., Mirzadeh S.,Brechbiel M. W. and Colcher D. (1989) Comparative biodistributionsof yttrium- and indium-labeled monoclonal antibody B72.3 in athymicmice bearing human colon carcinoma xenografts. J. Nucl. Med. 30,672–682.

12. Roy R., Katzenellenbogen E. and Jennings H. J. (1984) Improvedprocedures for the conjugation of oligosaccharides to protein by reduc-tive amination. Can. J. Biochem. Cell Biol. 62, 270–275.

13. Sakahara H., Endo K., Koizumi M., Nakashima T., Kunimatsu M.,Watanabe Y., Kawamura Y., Nakamura T., Tanaka H., Kotoura Y.,Yamamuro T., Hosoi S., Toyama S. and Torizuka K. (1988) Relation-

ship between in vitro binding activity and in vivo tumor accumulationof radiolabeled monoclonal antibodies. J. Nucl. Med. 29, 235–240.

14. Schlom J., Eggensperger D., Colcher D., Molinolo A., Houchens D.,Miller L. S., Hinkle G. and Siler K. (1992) Therapeutic advantage ofhigh-affinity anticarcinoma radioimmunoconjugates. Cancer Res. 52,1067–1072.

15. Schrappe M., Bumol T. F., Apelgren L. D., Briggs S. L., Koppel G. A.,Markowitz D. D., Mueller B. M. and Reisfeld R. A. (1992) Long-termgrowth suppression of human glioma xenografts by chemoimmunocon-jugates of 4-desacetylvinblastine-3-carboxyhydrazide and monoclonalantibody 9.2.27. Cancer Res. 52, 3838–3844.

16. Shih L. B., Sharkey R. M., Primus F. J. and Goldenberg D. M. (1988)Site-specific linkage of methotrexate to monoclonal antibodies using anintermediate carrier. Int. J. Cancer 41, 832–839.

17. Sivam G. P., Martin P. J., Reisfeld R. A. and Mueller B. M. (1995)Therapeutic efficacy of a doxorubicin immunoconjugate in a preclinicalmodel of spontaneous metastatic human melanoma. Cancer Res. 55,2352–2356.

18. Starling J. J., Maciak R. S., Law K. L., Hinson N. A., Briggs S. L.,Laguzza B. C. and Johnson D. A. (1991) In vivo antitumor activity of amonoclonal antibody-vinca alkaloid immunoconjugate directed againsta solid tumor membrane antigen characterized by heterogeneous expres-sion and noninternalization of antibody-antigen complexes. Cancer Res.51, 2965–2972.

19. Van den Abbeele A. D., Aaronson R. A., Adelstein S. J. and Kassis A. I.(1988) Does the in vitro testing of the immunoreactivity of an antibodyreflect its in vivo behavior? J. Nucl. Med. Allied Sci. 32, 260–267.

20. Wagner E., Cotten M., Mechtler K., Kirlappos H. and Birnstiel M. L.(1991) DNA-binding transferrin conjugates as functional gene-deliveryagents: Synthesis by linkage of polylysine or ethidium homodimer to thetransferrin carbohydrate moiety. Bioconjugate Chem. 2, 226–231.

21. Zalutsky M. R., Noska M. A., Colapinto E. V., Garg P. K. and BignerD. D. (1989) Enhanced tumor localization and in vivo stability of amonoclonal antibody radioiodinated using N-succinimidyl 3-(tri-n-butylstannyl)benzoate. Cancer Res. 49, 5543–5549.

278 R. S. Harapanhalli et al.